Skip to content

[Feature]: Optimize ROCm Attention Backend on V1 #66

@tjtanaa

Description

@tjtanaa

🚀 The feature, motivation and pitch

Motivation: To bridge the gap between V1 Engine vs V0 multi step = 10

Find out if the v1 AITER MHA kernel has the same FA CK kernel invocation as in the V0 Engine CK Flash Attention.
The reason is to find out if there is any difference in the shape that is causing the Backend to be so slow.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions